Stephen Hahn wrote:
* Shawn Walker <[email protected]> [2009-03-12 22:59]:
====================
Summary
====================
At the moment, users wanting to obtain publisher, mirror, or origin
information have to parse the printed output of the "pkg publisher"
command. This isn't locale safe and is prone to break whenever the
output format changes.
Since the output format was never intended as a stable interface for
consumers, an advanced set of listing options is desirable.
...
The common approach to this type of CLI interaction is -o
list_of_fields, as in ps(1), svcs(1), and others. Why is the
partition into types needed?
When a single property applies to multiple types, how else would a user
request retrieval of only those types? As I mentioned at the end of my
previous email, I was still debating whether -t must be combined with -p
or not. I could envision multiple ways that using -p would work by itself.
Part of the problem here is that the publisher command displays more
than just "publisher" information. It also displays repository,
repository mirror, and repository origin information.
So then, without partitioning into types, how would the user retrieve
only the URIs of origins?
Another thought I had was whether this should be a separate command,
such as "get-publisher" to fit better with CLIP and with our existing
"set-publisher" and "unset-publisher" commands.
I would also probably ask for a more strict parseable output format or
format(s), than the column-separated output we currently have.
A -f (--format) or -o (--output) option?
Cheers,
--
Shawn Walker
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
